Evidence That Friends Really Are The Family We Choose

07/14/2014 05:12pm ET | Updated July 14, 2014

Sister from another mister. Brother from another mother. The family you choose.

When it comes to describing your friends, those turns of phrase may be a lot more accurate than you think.

Genetics researchers at the University of California, San Diego and Yale University have found that friends can share a lot of the same genes -- the same amount of genes, in fact, as are shared by fourth cousins.

"In a lot of genetics research, the assumption is that we are interacting with lots of people who are not related to us, so genes aren't going to matter," said study researcher James Fowler, Ph.D., a medical genetics professor at UCSD. "This paper suggests that this is not the case at all; our friends are sort of like our family. It's quite extraordinary."

Besides providing geneticists with a whole new set of human relationships to examine (beyond blood relatives), the findings could "completely change our theory of evolution," Fowler told The Huffington Post. "Right now, it's based on the individual, and this work challenges that view; it suggests that we really can't understand evolution until we understand social networks."

Say for instance, that you're an ancient human who was born with a genetic mutation that allowed you to speak, Fowler explained. If you come across a friend that can also speak, you both have an advantage: You're better able to survive together, and more likely to reproduce and pass on your own genes than other people who don't have the ability to speak. Eventually, humans who can speak multiply and go on to dominate the community, Fowler hypothesized. It's that social network understanding of evolution, said Fowler, that could explain why some genes are changing faster than others.

"One of the things we found was that the genes that tend to be most similar [among friends] are also the genes that have been evolving the fastest," said Fowler. "That's really exciting, because it's supportive of a theory that one of the reasons human evolution has been speeding up over the last 30,000 years is because of network effects."

Fowler's study, co-written by Yale evolutionary biology professor Nicholas Christakis, M.D., Ph.D., MPH, was published online Monday in the journal Proceedings of the National Academy of Sciences.


Fowler and his team examined 1,932 subjects from the Framingham Heart Study, a well-known and much-researched cohort made up of three generations of Irish- and Italian-Americans from a small town outside of Boston.

They identified 1,300 pairs of friends among that group and compared their genetic markers -- about 1.5 million of them -- to each other. They also compared individuals' genes to strangers within the same group, to see if the friend pairs had more genes in common than the stranger pairs.

Indeed, researchers found that the friends did share a lot of the same genes -- as many as two people who share a great-great-great-grandparent.

"Most people don't even know who their fourth cousins are!" said Christakis in a statement. "Yet we are somehow, among a myriad of possibilities, managing to select as friends the people who resemble our kin."

At this point, it's important to note that the friends weren't actually fourth cousins. Fowler controlled for possible relatedness by removing any pairs of friends who may have even been slightly related.

Fowler found that the friend pairs were most likely to share the genes responsible for a sense of smell. It could be because smells can both draw and repel people from similar environments, making them more likely to find each other, said Fowler. Friends were also more likely to have very different immunity-related genes, which actually has the effect of giving someone extra immunity.

"The difference can be helpful for us because if they're not susceptible to a certain disease, they're not going to spread it to us," explained Fowler.

For the final phase of the experiment, researchers developed a genetic test to predict friendship among the study participants, and it worked. Fowler was able to predict friendship or stranger-ship among 450 participants just by examining their genes. It's the same method geneticists use to predict obesity, schizophrenia, and bipolar disorder in people.

One major drawback to the study: All of the participants were of European descent, drawn from a close-knit community that spans several generations. The makes any generalization about the wider population, and other ethnicities, difficult, Fowler admitted. On the other hand, Fowler said he was fairly certain people weren't simply choosing friends from the same ethnic group over friends with other ethnicities, simply because there were no other ethnicities to choose from.


But it's precisely the sample's racial and ancestral composition (or "population structure") that makes John Novembre, Ph.D., associate professor of Human Genetics at the University of Chicago, wary of Fowler's results.

"It's well-accepted that people tend to be friends with others of similar ancestry, but controlling for that effect is tricky and crucial for this study," wrote Novembre in an email to HuffPost.

Novembre wasn't involved with Fowler and Christakis' study, but he has conducted research on European genetic structure. His expertise makes him sensitive to potential pitfalls that can develop when researchers try to disentangle ethnicity, ancestry, social ties and genetics, he said. But even if Fowler and Christakis had controlled for population structure perfectly, said Novembre, the results are still "very subtle."

"If taken at face-value, that friends are on average as similar as fourth cousins, the effect is still very subtle," Novembre concluded. "I would guess any one person wouldn't be able to detect such a small effect acting in their life, especially relative to the much stronger societal factors that predict friendships."

Fowler admitted that Novembre's criticisms were fair. However, the subtle friendship effect in genes, he said, are about the same as genetic tests commonly used to predict medical outcomes. "I would also add that our out-of-sample test shows we can do prediction of friendship at the same accuracy as our best genetic tests for obesity, bipolar disorder, and schizophrenia," said Fowler in reaction to Novembre's perspective. "So the effects are indeed subtle, but no more so than other genetic studies of important biological and social outcomes."

Suggest a correction